Skill UI

A comprehensive guide to managing training data and model artifacts securely on cloud GPU instances (e.g., Vast.ai). It covers best practices for data transfer (SCP, compressed, cloud staging), implementing AES-256 encryption, managing checkpoints to S3, and performing secure cleanup before instance destruction, ensuring compliance and data integrity.

A comprehensive guide for deploying Machine Learning training jobs and inference services on the Vast.ai GPU cloud. This skill covers the entire deployment lifecycle, including optimizing Docker images for performance, creating automated provisioning scripts using the Vast.ai CLI, implementing robust data transfer strategies (SCP, rsync), and performing post-deployment health checks.

This guide details the production-grade reference architecture for running complex GPU compute workflows on Vast.ai. It covers the foundational three-tier pattern—Orchestrator, GPU Workers, and Artifact Storage—providing blueprints for job queue management, fault-tolerant training pipelines, and robust checkpointing. Use this when designing scalable and reliable ML/AI training systems.

Automates wandb experiment logger tasks within ML training, providing guidance, best practices, and code or configuration examples plus validation to keep data prep, training, hyperparameter tuning, and experiment tracking consistent.

PathML is a Python toolkit for computational pathology workflows, loading 160+ slide formats, preprocessing H&E images, constructing spatial graphs, training HoVer-Net/HACTNet models, and analyzing multiplex imaging for ML-ready pathology data.

Hardware-agnostic quantum ML framework enabling gradient-based training of quantum circuits, hybrid quantum-classical models, and seamless integration with PyTorch/JAX/TensorFlow alongside device portability across IBM/Google/Rigetti/IonQ.

PyTorch Lightning Toolkit streamlines PyTorch model training by organizing LightningModules, LightningDataModules, Trainer configs, callbacks, logging, and distributed strategies for scalable multi-GPU/TPU runs in research or production workflows.

This skill automates the end-to-end workflow for deploying and executing ML training and inference experiments. It intelligently detects the target environment (Local GPU, Remote SSH, Vast.ai, Modal serverless) and handles necessary steps like GPU pre-checks, code synchronization (rsync/git), integrating ML tracking tools (W&B), and launching the compute job via screen sessions or serverless deployment. Use when initiating a complex training job.

Trackio is a comprehensive library for logging, visualizing, and managing ML experiment metrics and alerts. It allows developers to use a Python API to log metrics and set diagnostic alerts (INFO/WARN/ERROR) during training. Furthermore, a dedicated CLI enables metric retrieval and dashboard viewing, syncing data to Hugging Face Spaces for persistent, remote monitoring, making it ideal for autonomous ML agent workflows.

Collection of battle-tested PyTorch training recipes covering architecture selection, optimizers, LR scheduling, precision tips, and debugging to train or fine-tune LLMs, vision, diffusion, and biomedical models with high GPU throughput.

A zero-config, serverless cloud platform designed for running intensive GPU workloads, including model training, fine-tuning, and inference. It eliminates the need for complex infrastructure setup (no SSH, no Docker) while supporting local-first development and auto-scaling for various high-performance GPUs (A100, H100). Ideal for data scientists and ML engineers requiring scalable and cost-efficient remote compute resources.

PathML is a comprehensive Python framework designed for computational pathology workflows. It supports end-to-end analysis of Whole Slide Images (WSI), including loading 160+ proprietary formats, performing advanced stain normalization and tissue segmentation, constructing spatial cell graphs, and training deep learning models (e.g., HoVer-Net) for multiplex immunofluorescence and spatial proteomics. Ideal for large-scale biomedical image analysis.